| abstract |
A method of calibrating an analytical instrument, for example an atomic absorption spectrophotometer, comprises measuring the absorbance of a plurality of standards of known concentration (100) and plotting the measured absorbance against concentration (101). A straight line is fitted to the plotted points (102) and a quality co-efficient calculated (103). If the quality co-efficient is acceptable (104) the calibration line is used for measurement of samples (105). If not, then the slope of the line joining each point to the origin is determined and if the slopes are random (107) then a robust regression technique is used to fit the calibration line (108). If outliers are then detected (109) it is determined which points are outliers (110) and appropriate action taken, for example to restrict the range if the last point(s) is/are outliers (111). If the slopes determined in step (106) are not random, then, provided more than 4 points remain (113), the slope of each point with respect to the first point is determined (114). If they are again not random (115), then a curved calibration line is diagnosed, while if they are random, a straight line not passing through the origin is diagnosed (117). In atomic absorption spectroscopy, a straight line not passing through the origin indicates a problem with the blank solution, for example, a problem of contamination. |